What happens if my tests don't pass during the Green phase?

The workflow will pause and notify you. The DEV agent must refine the implementation until all tests and linting checks pass before the green gate can be cleared.

How do I advance the workflow to the next step?

Once an agent completes its task, use the *prism-approve command to validate the current gate and move the workflow into the next phase (e.g., moving from RED tests to GREEN implementation).

Where is the current workflow progress stored?

State information is tracked in a local file at .claude/prism-loop.local.md, which includes the current step, active agent, and the path to the current story file.

What is the Ralph Wiggum pattern used in this skill?

It is a self-referential loop technique that allows Claude to maintain workflow context and automatically progress through complex multi-step development tasks without losing state.

Can I skip the TDD Red phase?

No, the workflow enforces TDD principles. The 'red gate' ensures that tests are written and verified as failing before implementation begins to maintain high code integrity.

PRISM TDD Workflow Loop

Name: PRISM TDD Workflow Loop
Author: resolve-io

byresolve-io

•

Security & Testing

Automates the end-to-end TDD development cycle by orchestrating AI agents through planning, testing, and implementation phases.

The PRISM Workflow Loop is a specialized orchestration skill that implements a rigorous Test-Driven Development (TDD) cycle using the Ralph Wiggum self-referential loop pattern. It coordinates multiple AI agents—Scrum Master (SM) for planning, Quality Assurance (QA) for testing, and Developer (DEV) for implementation—to ensure software is built with intention and resilience. By enforcing mandatory 'Red' and 'Green' gates, the skill prevents code from being written without failing tests and ensures all requirements are met before completion. This skill is ideal for developers following the PRISM methodology who want a predictable, automated framework for building maintainable software within Claude Code.

Key Features

01Integrated test execution and linting verification

021 GitHub stars

03Mandatory TDD validation gates (RED and GREEN states)

04Real-time state tracking and status reporting

05Automated agent orchestration (SM, QA, and DEV roles)

06Ralph Wiggum self-referential loop for autonomous progression

Use Cases

01Implementing new features using a strict Test-Driven Development approach

02Ensuring code quality through mandatory validation of failing and passing test states

03Automating the transition between requirements planning, test creation, and coding

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add resolve-io/.prism prism-loop

For use in Claude.ai and ChatGPT

Download Skill